Data Description

Description of population

The SEPAGES cohort includes 484 pregnant women. The main analysis is to be done on the continuous variables , that have the compound level in \(µg/L\) with filled-in values below detection limits (< LOD) and standarization for protocol variables. The pooled urine samples are taken during the second and the third trimester of pregnancy, so each woman has two measurements for each of the compounds. There are 477 samples corresponding to the second trimester while there are 454 women who have both the urine samples for the second and third trimester. 384 of the individuals who have the second trimester urine samples have all three thyroid hormone measurements (T4T, T3T and TSH) and do not have known thyroid problems (hypothyroidism, hyperthyroidism or goiter).

* includes individuals with oxoMINCH, ohMINCH or ohMPHP missing values.

Endocrine Disruptors

Continuous values

There are 48 standarized measures for each individual (24 for t2 and 24 for t3). The measures for almost all the compounds are very skewed to 0 while others like have very significant values. To better recognize the distribution of the compounds, we are going to use the log-transformed values.

Continuous Endocrine Disruptors
Compound Name Type P5% Median P95% N LOD (µgr/L) LOQ (µgr/L)
1 BPA Bisphenol A Phenol 0.5608 1.6117616 7.294 411 0.04 0.10
3 ETPA Ethyl paraben Phenol 0.2896 0.9534961 42.766 411 0.04 0.10
10 MEPA Methyl paraben Phenol 2.7339 12.8359869 225.812 411 0.04 0.10
17 OXBE Oxybenzone Phenol 0.1777 0.8144442 21.971 411 0.04 0.10
20 PRPA Propyl paraben Phenol 0.0050 0.4651963 68.248 411 0.04 0.10
21 TRCS Triclosan Phenol 0.2082 0.8531883 139.362 411 0.04 0.10
2 cxMiNP Mono-4-methyl-7-carboxyoctyl phthalate Phtalate 2.5603 4.6063000 25.419 411 0.40 1.00
4 MBzP Mono benzyl phthalate Phtalate 1.6958 4.7391368 15.519 411 0.07 0.20
5 MECPP Mono-2-ethyl 5-carboxypentyl phthalate Phtalate 5.0048 9.8975720 26.783 411 0.70 2.00
6 MEHHP Mono-2-ethyl-5-hydroxyhexyl phthalate Phtalate 3.2318 7.0804041 24.311 411 0.20 0.50
7 MEHP Mono-2-ethylhexyl phthalate Phtalate 0.7464 2.3519087 8.344 411 0.20 0.50
8 MEOHP Mono-2-ethyl-5-oxohexyl phthalate Phtalate 2.2695 5.0367465 17.173 411 0.20 0.50
9 MEP Monoethyl phthalate Phtalate 6.3769 24.0060508 123.292 411 0.20 0.50
11 MiBP Mono-iso-butyl phthalate Phtalate 6.8283 15.7095278 45.452 411 0.20 0.50
12 MMCHP Mono-2-carboxymethyl hexyl phthalate Phtalate 4.1872 7.6656170 20.094 411 0.70 2.00
13 MnBP Mono-n-butyl phthalate Phtalate 4.8965 11.2218365 32.903 411 0.20 0.50
14 ohMINCH 2-(((Hydroxy-4-methyloctyl)oxy)carbonyl)cyclohexanecarboxylic Acid Phtalate 0.7429 1.7789841 17.074 402 0.07 0.20
15 ohMiNP Mono-4-methyl-7-hydroxyoctyl phthalate Phtalate 1.6723 4.6921000 27.968 411 0.10 0.25
16 ohMPHP 6-Hydroxy Monopropylheptylphthalate Phtalate 0.4973 0.8611698 2.614 402 0.07 0.20
18 oxoMINCH 2-(((4-Methyl-7-oxyooctyl)oxy)carbonyl)cyclohexanecarboxylic Acid Phtalate 0.5719 1.3747216 11.632 402 0.07 0.20
19 oxoMiNP Mono-4-methyl-7-oxooctyl phthalate Phtalate 0.8191 2.1235963 12.444 411 0.10 0.25
Categorical Endocrine Disruptors
Compound Name Type Under LOD Between LOD and LOQ Over LOQ N NA’s LOD (µgr/L) LOQ (µgr/L)
BPS Bisphenol S Phenol 306 19 86 411 0 0.10 0.40
BUPA Butyl paraben Phenol 309 57 45 411 0 0.07 0.20
BPF Bisphenol F Phenol 405 0 6 411 0 0.07 0.20
TRCB Triclocarban Phenol 409 1 1 411 0 0.04 0.10
BPAF Bisphenol AF Phenol 411 0 0 411 0 0.02 0.05
BPB Bisphenol B Phenol 411 0 0 411 0 0.03 0.10

Most of these endocrine disruptors are under the LOD and cannot be detected. The only ones that have significant number of samples over the LOQ are BPS and BUPA and are the only ones we recommend to keep.

Hierarchical Cluster Analysis

When we study the correlation for the measurements of the second trimester, there is a high correlation between the metabolites of DEHP (MEOHP, MECPP, MEHP, MMCHP and MEHHP). The same is true for the metabolites of DiNP (ohMiNP, oxoMiNP and cxMiNP) and the metabolites of DINCH (oxo-MINCH and oh-MINCH). Some phenols are also correlated between each other, such as MEP and ETPA, as well as PRPA and MEPA.

Thyroid Hormones

Thyroid hormone samples for individuals with endocrine disruptor samples during the second trimester of the pregnancy and no thyroid problems
Hormone P5% Median P95% N
Free T3 1.70 2.10 2.600 411
Total T3 0.90 1.20 1.600 384
Free T4 5.40 7.20 10.350 411
Total T4 75.75 95.20 121.850 411
TSH 0.63 1.29 2.545 411

For the 437 women who do not have thyroid problems and have the second trimester endocrine disruptor samples, there are 411 samples for Total T4 (TT4) and TSH and, 384 samples for Total T3 (TT3). Only TT4 concentration appears to approach a normal distribution. This is not the case for TT3 or TSH (Appendix).

Correlation between thyroid hormones

Thyroid hormones are moderately correlated beween each other. The total forms of T3 and T4 are slightly correlated between each other and, TT3 an FT3 have a similar correlation with each other. The correlation between TSH and all the other hormones is very close to 0.

Thyroid hormones and endocrine disruptors

Confounder Variables

Continuous confounders
Confounder P5% Median P95% N
Mother Age (years) 26.6160 32.07118 39.003 411
Mother BMI before pregnancy 18.2754 21.48438 28.644 407
Gestational age at serum sampling (weeks) 16.0000 19.14286 27.000 411
Iodine (mg/L) 31.1984 87.49755 264.460 394
Selenium (µg/L) 0.7900 0.96000 1.200 346

The mothers have a median age of 32.0711823, which is older than the average age of primiparous mothers in France, 30.7 years old (INSEE, 2019). Around 20% of the women had the serum sampling much later than the others. This group will not be used for the main analysis but for the latter sensitivity analysis.

Categorical Confounders
N (%)
BMI
Underweight 26 6.3%
Normal 307 74.7%
Overweight 59 14.4%
Obese 15 3.6%
NA 4 1.0%
Parity
Nulliparous 189 46.0%
Primiparous 181 44.0%
Multiparous 40 9.7%
NA 1 0.2%
Education
High School or less 69 17%
Undergrad 110 27%
Postgrad 230 56%
NA 2 0%

46.0% of the mothers included in the cohort were nulliparous before the currently studied offspring. A similar percentage was primiparous (44.0%) while only 9.7% had already given birth to 2 or more offsprings. 56% of the participants have a postgraduate degree, 27% have an undergraduate degree and 17% have a High school diploma or less.

Protocol Variables

Thyroid hormone batch

Number of samples per batch of the thryoid hormones
1 2 3 4 5 6 7
FT3 39 71 78 74 80 69 39
FT4 39 71 78 74 80 29 40
TT3, TT4 and TSH 39 71 78 74 80 69 39

Iodine

Number of samples per batch of iodine
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 NA’s
Iodine 33 9 24 20 28 17 28 19 27 35 27 20 26 19 17 18 27 17

Time delay between thyroid hormone sampling and freezing

   Min. 1st Qu.  Median    Mean 3rd Qu.    Max.    NA's 
0.02014 0.07083 0.12951 0.35327 0.63889 3.68056       1 

Thyroid Hormones and Confounder Variables

Some of the thyroid hormones have a slight correlation with the confounder variables. The positive correlation between the BMI and the TT3 is the highest (\(r=\) 0.2392242 ). The gestational age at serum sampling has a positive correlation with FT4 (\(r=\) -0.2314729 ) and with FT3 (\(r=\) 0.2392242 ). In addition, the mother’s age has a negative correlation with TT4 (\(r=\) -0.1716068 ).

P-value of significance tests for the covariates for every thyroid hormone
Iodinea Seleniuma Freezing.Timea Batchb
TT3 0.1427322 0.01123 0.4628950 0.000203
TT4 0.9192887 0.8354 0.6434830 2.193e-09
TSH 0.6764185 0.7108 0.7250339 0.3002
a Univariate Linear regression
b Analysis of variance test

For the second trimester samples on women without thyroid problems, we studied the significance of univariate regression tests on the four possible confounders for TT3, TT4 and TSH. TT3 has a significant slope on the univariate linear regression on selenium and there is a significant effect of batch on TT3 and TT4, all of which have a p-value under the chosen threshold of 0.2. When this significance tests are done on the entire population, including the women with thyroid problems, iodine concentration also has a significant effect on TT4.

When we study the linear regression of iodine concentration on each of the three thyroid hormones (TT3, TT4 and TSH), the slope has a p-value inferior to the threshold of 0.2 for TT3 and TT4. These results align with Ahad and Ganie. * DOROTHY: : According to Ahad and Ganie, iodine uptake into the follicular cells could be considered the first step in synthesis of thyroid hormones (T3 and T4) (Ahad and Ganie, 2010). The inclusion of iodine in the model depends on which of the three relationship models we consider.

The batch in which the thyroid hormones were measured appears to have a significant effect on the measurement for TT3 (p-value=2.0310^{-4}) and for TT4 (p-value=2.1910^{-9}) but not for TSH (p-value=0.3002).

2020-03-12